In [1]:
import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt
#from google.colab import drive
#drive.mount('/content/drive')
In [2]:
train_df = pd.read_csv("train_data.csv", sep=",", na_values=[" ","?","#"])
test_df = pd.read_csv("test_data.csv", sep=",", na_values=[" ","?","#"])
discount_df = pd.read_csv("discount_features.csv", sep=",", na_values=[" ","?","#"]
,parse_dates=['date'])
foot_fall_df = pd.read_csv("foot_fall.csv", sep=",", na_values=[" ","?","#"])
In [3]:
print(train_df.shape)
print(test_df.shape)
print(discount_df.shape)
print(foot_fall_df.shape)
In [4]:
train_df.head()
Out[4]:
In [5]:
train_df.tail()
Out[5]:
Observations¶
The train data is dated from 2nd january 2015 to 30th June 2018
In [6]:
train_df.describe()
Out[6]:
In [7]:
len(train_df[train_df["sales"]<0])
Out[7]:
Observation¶
1672 rows have negative sales
In [8]:
test_df.head()
Out[8]:
In [9]:
test_df.tail()
Out[9]:
Observations¶
The test data ranges from 1st July 2018 to 31st July 2018
In [10]:
discount_df.head()
Out[10]:
In [11]:
discount_df.tail()
Out[11]:
Observation¶
We have discount data from 1 April 2016 to 31 July 2018
In [12]:
foot_fall_df.head(10)
Out[12]:
In [13]:
train_df["sales"].abs().min()
Out[13]:
Observation¶
In [14]:
## Checking for NA values
In [15]:
train_df.isna().sum()
Out[15]:
In [16]:
test_df.isna().sum()
Out[16]:
In [17]:
foot_fall_df.isna().sum()
Out[17]:
In [18]:
discount_df.isna().sum()
Out[18]:
Observations¶
Initially there are no na values in the data
In [19]:
discount_df.dtypes
Out[19]:
In [20]:
## Creating Copies of the datasets
train_df_copy = train_df.copy()
test_df_copy = test_df.copy()
discount_df_copy = discount_df.copy()
foot_fall_df_copy = foot_fall_df.copy()
In [21]:
## dismantling the date column in discount_features_copy into year, month and day
discount_df_copy.dtypes
Out[21]:
In [22]:
discount_df_copy['year'] = discount_df_copy['date'].dt.year
discount_df_copy['month'] = discount_df_copy['date'].dt.month
discount_df_copy['day'] = discount_df_copy['date'].dt.day
In [23]:
discount_df_copy.head()
Out[23]:
In [24]:
train_df_discounted_with_date = pd.merge(train_df_copy, discount_df_copy
,how="inner", on=["year","month","day","city","medicine"])
train_df_discounted_with_date.head()
Out[24]:
In [25]:
## Dropping the date column in discount_features_copy
discount_df_copy.drop("date", axis = 1, inplace=True)
In [26]:
discount_df_copy.head()
Out[26]:
In [27]:
## Merging the train_data_copy and the discounted_data_copy for adding discount feature
train_df_discounted = pd.merge(train_df_copy, discount_df_copy
,how="left", on=["year","month","day","city","medicine"])
In [28]:
print(train_df_copy.shape)
print(train_df_discounted.shape)
In [29]:
##train_discounted.drop_duplicates(inplace=True)
In [30]:
print(train_df_copy.shape)
print(train_df_discounted.shape)
In [31]:
train_df_discounted.head()
Out[31]:
In [32]:
train_df_discounted.tail()
Out[32]:
In [33]:
train_df_discounted.isna().sum()
Out[33]:
In [34]:
train_df_discounted.fillna(0,inplace=True)
In [35]:
train_df_discounted.head()
Out[35]:
In [36]:
## Dropping id column from the test_data_copy()
test_df_copy.drop("id", axis=1, inplace=True)
In [37]:
test_df_copy.head()
Out[37]:
In [38]:
## Merging and test and discounted
test_df_discounted = pd.merge(test_df_copy, discount_df_copy, how="inner"
,on=["year","month","day","city","medicine"])
In [39]:
test_df_discounted.head()
Out[39]:
In [40]:
print(test_df_copy.shape)
print(test_df_discounted.shape)
In [41]:
print(train_df_discounted.medicine.nunique())
print(test_df_discounted.medicine.nunique())
In [42]:
train_neg_df_discounted = train_df_discounted[train_df_discounted["sales"]<0]
train_neg_df_discounted.shape
Out[42]:
In [43]:
train_neg_df_discounted.head()
Out[43]:
In [44]:
train_neg_df_discounted.isna().sum()
Out[44]:
In [45]:
train_neg_df_discounted.medicine.nunique()
Out[45]:
In [46]:
train_df_discounted.drop(train_df_discounted[train_df_discounted["sales"]<0].index,axis=0,inplace=True)
In [47]:
train_df_discounted.shape
Out[47]:
In [48]:
train_df_discounted.dtypes
Out[48]:
In [49]:
cat_columns = ["year","month","day","city","medicine","discounted"]
def change_to_category(df,col_list):
for col in col_list:
df[col] = df[col].astype("category")
return df
In [50]:
train_discounted = change_to_category(train_df_discounted,cat_columns)
In [51]:
train_df_discounted.dtypes
Out[51]:
In [52]:
## Creating City wise datasets
city_frame = {}
for city in train_df_discounted.city.unique():
df = train_df_discounted[train_df_discounted["city"]==city]
city_frame[city] = df
In [53]:
kolkata_df = city_frame[1]
other_df = city_frame[2]
jaipur_df = city_frame[3]
mumbai_df = city_frame[4]
delhi_df = city_frame[5]
chennai_df = city_frame[6]
bengaluru_df = city_frame[7]
hyderabad_df = city_frame[8]
ahemdabad_df = city_frame[9]
lucknow_df = city_frame[10]
In [54]:
print(kolkata_df.shape)
print(other_df.shape)
print(jaipur_df.shape)
print(mumbai_df.shape)
print(delhi_df.shape)
print(chennai_df.shape)
print(bengaluru_df.shape)
print(hyderabad_df.shape)
print(ahemdabad_df.shape)
print(lucknow_df.shape)
Basic Exploratory Data Analysis¶
In [55]:
## Yearly analysis
yearly_sales_df = train_df_discounted.groupby(["year"]).agg({"sales":sum}).reset_index()
yearly_sales_df = yearly_sales_df.sort_values(by="sales",ascending=False)
yearly_sales_df
Out[55]:
In [56]:
ax = sns.barplot(x="sales", y="year", data=yearly_sales_df)
Observations¶
1) Overall 2017 had most no of sales 2) Overall 2018 had least no of sales
In [57]:
## Monthly analysis
monthly_sales_df = train_df_discounted.groupby(["month"]).agg({"sales":sum}).reset_index()
monthly_sales_df = monthly_sales_df.sort_values(by="sales", ascending=False)
monthly_sales_df
Out[57]:
In [58]:
ax = sns.barplot(x="month", y="sales", data = monthly_sales_df)
Observations¶
March has the highest sales followed by January and December. August has the lowest sales
City-wise Analysis¶
In [59]:
train_df_discounted_copy_city = train_df_discounted.copy()
city = {4: 'Mumbai', 5: 'Delhi', 7: 'Bangalore', 1: 'Kolkata',
6: 'Chennai', 8: 'Hyderabad', 9: 'Ahmedabad', 3: 'Jaipur', 10: 'Lucknow', 2: 'Other'}
train_df_discounted_copy_city["city"]=train_df_discounted_copy_city["city"].map(city)
train_df_discounted_copy_city.head()
Out[59]:
In [60]:
import matplotlib.pyplot as plt
import seaborn as sns
def do_city_analysis(df, city):
city_df = df[df["city"] == city]
city_df["year"] = city_df["year"].astype("category")
city_df["month"] = city_df["month"].astype("category")
city_df["day"] = city_df["day"].astype("category")
fig_dims = (20, 20)
f, axes = plt.subplots(nrows=3, ncols=1, figsize = fig_dims)
yearly_sales_df = city_df.groupby(["year"]).agg({"sales":sum}).reset_index()
yearly_sales_df = yearly_sales_df.sort_values(by="sales", ascending=False)
sns.barplot(x="sales", y="year", data=yearly_sales_df, ax=axes[0])
monthy_sales_df = city_df.groupby(["month"]).agg({"sales":sum}).reset_index()
monthy_sales_df = monthy_sales_df.sort_values(by="sales", ascending=False)
sns.barplot(x="sales", y="month", data=monthy_sales_df, ax=axes[1])
daily_sales_df = city_df.groupby(['day']).agg({"sales":sum}).reset_index()
daily_sales_df = daily_sales_df.sort_values(by="sales", ascending = False)
sns.barplot(x="sales", y="day", data=daily_sales_df, ax=axes[2])
medicine_df = city_df.groupby(["medicine"]).agg({"sales":sum}).reset_index()
medicine_df = medicine_df.sort_values(by="sales", ascending=False)
zero_sales_meds = medicine_df[medicine_df["sales"]==0]["medicine"]
print(f"total # of medicenes with zero sale: {len(zero_sales_meds)}")
plt.show()
In [61]:
import matplotlib.pyplot as plt
def plot_top_meds(df, city, year, top=10):
df[(df["city"] == city) & (df["year"] == year)].groupby(["medicine"])["sales"].sum().sort_values(ascending = False)[:10].plot.bar()
plt.xticks(rotation = 45, ha = "right")
plt.title( f"top {top} medicines sold in {city} in {year}")
plt.ylabel("sales")
plt.show()
Kolkata Analysis¶
In [62]:
do_city_analysis(train_df_discounted_copy_city,"Kolkata")
Observations¶
- 2017 shows highest sales for kolkata
- 2015 shows lowest sales for kolkata
- For kolkata March has the highest sales followed by january and july and august has the lowest sales
- In Kolkata third day of the month has the most sales followed by second and the fifth day. 31st day has the lowest sales
Kolkata Medicine Analysis¶
In [63]:
# Top 10 medicines in 2015 in Kolkata
plot_top_meds(train_df_discounted_copy_city, "Kolkata",2015)
In [64]:
# Top 10 medicines in 2016 in Kolkata
plot_top_meds(train_df_discounted_copy_city, "Kolkata",2016)
In [65]:
# Top 10 medicines in 2017 in Kolkata
plot_top_meds(train_df_discounted_copy_city, "Kolkata",2017)
In [66]:
# Top 10 medicines in 2018 in Kolkata
plot_top_meds(train_df_discounted_copy_city, "Kolkata",2018)
In [67]:
kolkata_medicine_df = kolkata_df.groupby(["medicine"]).agg({"sales":sum}).reset_index()
kolkata_medicine_df = kolkata_medicine_df.sort_values(by="sales", ascending=False)
kolkata_zero_sales_meds = kolkata_medicine_df[kolkata_medicine_df["sales"]==0]["medicine"]
len(kolkata_zero_sales_meds)
Out[67]:
Observations¶
- 1077 is the most sold medicine in kolkata in 2015
- 1077 is the most sold medicine in kolkata in 2016
- 1077 is the most sold medicine in kolkata in 2017
- 2103 is the most sold medicine in kolkata in 2018
- There are 122 medicines which have zero sales in kolkata
Top medicine discount analysis¶
In [215]:
kolkata_1077 = kolkata_df[kolkata_df["medicine"]==1077]
kolkata_1077_discount = kolkata_1077.groupby(["discounted"]).agg({"sales":sum}).reset_index()
kolkata_1077_discount
Out[215]:
discount had no role to play in the sales of 1077
Jaipur Analysis¶
In [68]:
do_city_analysis(train_df_discounted_copy_city, "Jaipur")
Observations¶
- For Jaipur 2017 has the highest sales followed by 2016 and 2015 has the lowest sales
- For Jaipur march has most sales followed by january and June. August has the least sales
- For Jaipur Day 3 has the highest sales followed by day 20 and 5. Day 31 has the least sales
Jaipur Medicine Analysis¶
In [69]:
# top 10 medicine in Jaipur in 2015
plot_top_meds(train_df_discounted_copy_city, city="Jaipur", year=2015)
In [70]:
# top 10 medicine in Jaipur in 2016
plot_top_meds(train_df_discounted_copy_city, city="Jaipur", year=2016)
In [71]:
# top 10 medicine in Jaipur in 2017
plot_top_meds(train_df_discounted_copy_city, city="Jaipur", year=2017)
In [72]:
# top 10 medicine in Jaipur in 2018
plot_top_meds(train_df_discounted_copy_city, city="Jaipur", year=2018)
In [73]:
jaipur_medicine_df = jaipur_df.groupby(["medicine"]).agg({"sales":sum}).reset_index()
jaipur_medicine_df = jaipur_medicine_df.sort_values(by="sales", ascending=False)
jaipur_zero_sales_meds = jaipur_medicine_df[jaipur_medicine_df["sales"]==0]
jaipur_zero_sales_meds = jaipur_zero_sales_meds["medicine"]
len(jaipur_zero_sales_meds)
Out[73]:
Observation¶
- In Jaipur medicine 1077 has the highest sales in 2015, 2017.
- In Jaipur medicine 2103 has the highest sales in 2016, 2018.
- There are 138 medicines in Jaipur that have 0 sales
Affect of Discount on top medicine in Jaipur¶
In [216]:
jaipur_1077 = jaipur_df[jaipur_df["medicine"]==1077]
jaipur_1077_discount = jaipur_1077.groupby(["discounted"]).agg({"sales":sum}).reset_index()
jaipur_1077_discount
Out[216]:
In [217]:
jaipur_2103 = jaipur_df[jaipur_df["medicine"]==2103]
jaipur_2103_discount = jaipur_2103.groupby(["discounted"]).agg({"sales":sum}).reset_index()
jaipur_2103_discount
Out[217]:
discount had no role to play in the sales of 1077 in jaipur discount had a role to play in the sales of 2103 in jaipur
Mumbai Analysis¶
In [74]:
do_city_analysis(train_df_discounted_copy_city, "Mumbai")
Observations¶
- For Mumbai 2017 has most sales followed by 2016 and 2018 has the least sales
- For Mumbai March has most sales followed by january, June and December. August has the least sales
- For Mumbai day 3 has the highest sales followed by day 2 and 1. Day 31 has the least sales
Mumbai Medicine Analysis¶
In [75]:
plot_top_meds(train_df_discounted_copy_city, city="Mumbai", year=2015)
In [76]:
plot_top_meds(train_df_discounted_copy_city, city="Mumbai", year=2016)
In [77]:
plot_top_meds(train_df_discounted_copy_city, city="Mumbai", year=2017)
In [78]:
plot_top_meds(train_df_discounted_copy_city, city="Mumbai", year=2018)
In [79]:
mumbai_medicine_df = mumbai_df.groupby(["medicine"]).agg({"sales":sum}).reset_index()
mumbai_medicine_df = mumbai_medicine_df.sort_values(by="sales", ascending=False)
mumbai_zero_sales_meds = mumbai_medicine_df[mumbai_medicine_df["sales"]==0]
mumbai_zero_sales_meds = mumbai_zero_sales_meds["medicine"]
len(mumbai_zero_sales_meds)
Out[79]:
Observations¶
- For Mumbai medicine type 1077 has the most sales in 2015
- For Mumbai medicine type 1077 has the most sales in 2016
- For Mumbai medicine type 1077 has the most sales in 2017
- For Mumbai medicine type 2103 has the most sales in 2018
- There are 119 medicines that have zero sales
Affect of Discount on Top medicine in Mumbai¶
In [218]:
mumbai_1077 = mumbai_df[mumbai_df["medicine"]==1077]
mumbai_1077_discount = mumbai_1077.groupby(["discounted"]).agg({"sales":sum}).reset_index()
mumbai_1077_discount
Out[218]:
In [219]:
mumbai_2103 = mumbai_df[mumbai_df["medicine"]==2103]
mumbai_2103_discount = mumbai_2103.groupby(["discounted"]).agg({"sales":sum}).reset_index()
mumbai_2103_discount
Out[219]:
Delhi Analysis¶
In [80]:
do_city_analysis(train_df_discounted_copy_city, "Delhi")
Observations¶
- For Delhi 2017 has most sales followed by 2016 and 2015 has the least sales
- March has the most sales followed by jan, june and Dec. Aug has the least sales
- For Delhi Day 2 has the most no of sales followed by day 3 and 1. Day 31 has the least sales
Delhi Medicine Analysis¶
In [81]:
plot_top_meds(train_df_discounted_copy_city, "Delhi", 2015)
In [82]:
plot_top_meds(train_df_discounted_copy_city, "Delhi", 2016)
In [83]:
plot_top_meds(train_df_discounted_copy_city, "Delhi", 2016)
In [84]:
plot_top_meds(train_df_discounted_copy_city, "Delhi", 2017)
In [85]:
plot_top_meds(train_df_discounted_copy_city, "Delhi", 2018)
In [86]:
delhi_medicine_df = delhi_df.groupby(["medicine"]).agg({"sales":sum}).reset_index()
delhi_medicine_df = delhi_medicine_df.sort_values(by="sales",ascending=False)
delhi_zero_sales_meds = delhi_medicine_df[delhi_medicine_df["sales"]==0]
delhi_zero_sales_meds = delhi_zero_sales_meds["medicine"]
len(delhi_zero_sales_meds)
Out[86]:
Observations¶
- For Delhi medicine type 574 had the most sales in 2015
- For Delhi medicine type 2103 had the most sales in 2016
- For Delhi medicine type 2103 had the most sales in 2017
- For Delhi medicine type 2103 had the most sales in 2018
- There are 113 medicines which have zero sales in Delhi
In [220]:
delhi_2103 = delhi_df[delhi_df["medicine"]==2103]
delhi_2103_discount = delhi_2103.groupby(["discounted"]).agg({"sales":sum}).reset_index()
delhi_2103_discount
Out[220]:
In [221]:
delhi_574 = delhi_df[delhi_df["medicine"]==574]
delh_574_discount = delhi_574.groupby(["discounted"]).agg({"sales":sum}).reset_index()
delh_574_discount
Out[221]:
discount have affect on sales of medicine 574 and 2103 in delhi
Chennai Analysis¶
In [87]:
do_city_analysis(train_df_discounted_copy_city, "Chennai")
Observation¶
- For Chennai 2017 has most sale followed by 2016 and 2018 has least no of sales
- For Chennai March has the most no of sales followed by dec, jan and may. Aug has the least sales
- for Chennai day 2 has most of the sales followed by day 3, 1 and 5. This implies most of the sale happend during first week of the month
Chennai Medicine Analysis¶
In [88]:
plot_top_meds(train_df_discounted_copy_city, "Chennai", 2015)
In [89]:
plot_top_meds(train_df_discounted_copy_city, "Chennai", 2016)
In [90]:
plot_top_meds(train_df_discounted_copy_city, "Chennai", 2017)
In [91]:
plot_top_meds(train_df_discounted_copy_city, "Chennai", 2018)
In [92]:
chennai_medicine_df = chennai_df.groupby(["medicine"]).agg({"sales":sum}).reset_index()
chennai_medicine_df = chennai_medicine_df.sort_values(by="sales", ascending=False)
chennai_zero_sales_meds = chennai_medicine_df[chennai_medicine_df["sales"]==0]
chennai_zero_sales_meds = chennai_zero_sales_meds["medicine"]
len(chennai_zero_sales_meds)
Out[92]:
Observation¶
- For Chennai medicine type 574 has most sales in 2015
- For Chennai medicine type 574 has most sales in 2016
- For Chennai medicine type 1517 has most sales in 2017
- For Chennai medicine type 1971 has most sales in 2018
- For Chennai there are 136 medicine that have zero sales
Affect of discount on top medicine in Chennai¶
In [222]:
chennai_574 = chennai_df[chennai_df["medicine"]==574]
chennai_574_discount = chennai_574.groupby(["discounted"]).agg({"sales":sum}).reset_index()
chennai_574_discount
Out[222]:
In [223]:
chennai_1517 = chennai_df[chennai_df["medicine"]==1517]
chennai_1517_discount = chennai_1517.groupby(["discounted"]).agg({"sales":sum}).reset_index()
chennai_1517_discount
Out[223]:
In [224]:
chennai_1971 = chennai_df[chennai_df["medicine"]==1971]
chennai_1971_discount = chennai_1971.groupby(["discounted"]).agg({"sales":sum}).reset_index()
chennai_1971_discount
Out[224]:
- discount has affect on the sales of 574 in chennai
- discount has no affect on the sales of 1971, 1517 in chennai
Bengaluru Analysis¶
In [93]:
do_city_analysis(train_df_discounted_copy_city, "Bangalore")
Observations¶
- For Bengaluru 2017 has most sale followed by 2016 and 2018 has least sales
- For Bengaluru March had most no of sales followed by jan, Dec and June. Aug has least no of sales
- For Bengaluru Day 3 has most no of sale followed by day 2, 1 and 5. This implies most sales happen during the first week of the month
Bengaluru Medicine Analysis¶
In [94]:
plot_top_meds(train_df_discounted_copy_city, "Bangalore", 2015)
In [95]:
plot_top_meds(train_df_discounted_copy_city, "Bangalore", 2016)
In [96]:
plot_top_meds(train_df_discounted_copy_city, "Bangalore", 2017)
In [97]:
plot_top_meds(train_df_discounted_copy_city, "Bangalore", 2018)
In [98]:
bengaluru_medicine_df = bengaluru_df.groupby(["medicine"]).agg({"sales":sum}).reset_index()
bengaluru_medicine_df = bengaluru_medicine_df.sort_values(by="sales", ascending= False)
bengaluru_zero_sales_meds = bengaluru_medicine_df[bengaluru_medicine_df["sales"]==0]
bengaluru_zero_sales_meds =bengaluru_zero_sales_meds["medicine"]
len(bengaluru_zero_sales_meds)
Out[98]:
In [99]:
bengaluru_medicine_df.head()
Out[99]:
Observations¶
- For Bengaluru medicine type 768 has most sales in 2015
- For Bengaluru medicine type 1517 has most sales in 2016
- For Bengaluru medicine type 1077 has most sales in 2017
- For Bengaluru medicine type 2103 has most sales in 2018
- For Bengaluru there are 115 medicines that have zero sales
In [225]:
bengaluru_768 = bengaluru_df[bengaluru_df["medicine"]==768]
bengaluru_768_discount = bengaluru_768.groupby(["discounted"]).agg({"sales":sum}).reset_index()
bengaluru_768_discount
Out[225]:
In [226]:
bengaluru_1517 = bengaluru_df[bengaluru_df["medicine"]==1517]
bengaluru_1517_discount = bengaluru_1517.groupby(["discounted"]).agg({"sales":sum}).reset_index()
bengaluru_1517_discount
Out[226]:
In [227]:
bengaluru_1077 = bengaluru_df[bengaluru_df["medicine"]==1077]
bengaluru_1077_discount = bengaluru_1077.groupby(["discounted"]).agg({"sales":sum}).reset_index()
bengaluru_1077_discount
Out[227]:
In [228]:
bengaluru_2103 = bengaluru_df[bengaluru_df["medicine"]==2103]
bengaluru_2103_discount = bengaluru_2103.groupby(["discounted"]).agg({"sales":sum}).reset_index()
bengaluru_2103_discount
Out[228]:
- discount has no affect on sales for medicine 768, 1517, 1077 in bengaluru
- discount has affect on sales for medicine 2103 in bengaluru
Hyderabad Analysis¶
In [100]:
do_city_analysis(train_df_discounted_copy_city, "Hyderabad")
Observations¶
- For Hyderabad 2017 show most sales followed by 2016 and 2018 has least sales
- For Hyderbad Dec has most no of sales followed by march, jan and May. Aug has least sales
- For Hyderabad Day 1 has the most sale followed by day 3 and 2 and day 31 has the least sales
Hyderabad Medicine Analysis¶
In [101]:
plot_top_meds(train_df_discounted_copy_city, "Hyderabad", 2015)
In [102]:
plot_top_meds(train_df_discounted_copy_city, "Hyderabad", 2016)
In [103]:
plot_top_meds(train_df_discounted_copy_city, "Hyderabad", 2017)
In [104]:
plot_top_meds(train_df_discounted_copy_city, "Hyderabad", 2018)
In [105]:
hyderabad_medicine_df = hyderabad_df.groupby(["medicine"]).agg({"sales":sum}).reset_index()
hyderabad_medicine_df = hyderabad_medicine_df.sort_values(by="sales", ascending= False)
hyderabad_zero_sales_meds = hyderabad_medicine_df[hyderabad_medicine_df["sales"]==0]
hyderabad_zero_sales_meds = hyderabad_zero_sales_meds["medicine"]
len(hyderabad_zero_sales_meds)
Out[105]:
In [106]:
hyderabad_medicine_df.head()
Out[106]:
Observations¶
- For Hyderabad medicine type 574 had most sales in 2015
- For Hyderabad medicine type 574 had most sales in 2016
- For Hyderabad medicine type 574 had most sales in 2017
- For Hyderabad medicine type 1971 had most sales in 2018
- For Hyderabd there were total 152 medicines that had zero sales
In [229]:
hyderabad_574 = hyderabad_df[hyderabad_df["medicine"]==574]
hyderabad_574_discount = hyderabad_574.groupby(["discounted"]).agg({"sales":sum}).reset_index()
hyderabad_574_discount
Out[229]:
In [230]:
hyderabad_1971 = hyderabad_df[hyderabad_df["medicine"]==1971]
hyderabad_1971_discount = hyderabad_1971.groupby(["discounted"]).agg({"sales":sum}).reset_index()
hyderabad_1971_discount
Out[230]:
- discount has affect on sales of medicine 574 in hyderabad
- discount has no affect on sales of medicine 1971 in hyderabad
Ahemdabad Analysis¶
In [107]:
do_city_analysis(train_df_discounted_copy_city, "Ahmedabad")
Observations¶
- for Ahemdabad 2017 had most sales followed by 2016 and 2015 had least sales
- For Ahemdabad December has most sales followed by March, June and Jan. Aug has the least sale
- For Ahemdabad day 3 had most sales followed by day 2, 1 and 6 and day 31 has least sales
Ahemdabad Medicine Analysis¶
In [108]:
plot_top_meds(train_df_discounted_copy_city, "Ahmedabad", 2015)
In [109]:
plot_top_meds(train_df_discounted_copy_city, "Ahmedabad", 2016)
In [110]:
plot_top_meds(train_df_discounted_copy_city, "Ahmedabad", 2017)
In [111]:
plot_top_meds(train_df_discounted_copy_city, "Ahmedabad", 2018)
In [112]:
ahemdabad_medicine_df = ahemdabad_df.groupby(["medicine"]).agg({"sales":sum}).reset_index()
ahemdabad_medicine_df = ahemdabad_medicine_df.sort_values(by="sales",ascending= False)
ahemdabad_zero_sales_meds = ahemdabad_medicine_df[ahemdabad_medicine_df["sales"]==0]
ahemdabad_zero_sales_meds = ahemdabad_zero_sales_meds["medicine"]
len(ahemdabad_zero_sales_meds)
Out[112]:
In [113]:
ahemdabad_medicine_df.head()
Out[113]:
Observations¶
- For Ahmedabad medicine type 316 had most sales in 2015
- For Ahmedabad medicine type 2103 had most sales in 2016
- For Ahmedabad medicine type 2103 had most sales in 2017
- For Ahmedabad medicine type 2103 had most sales in 2018
- For Ahmedabad there are 128 medicines that had zero sales
In [231]:
ahemdabad_316 = ahemdabad_df[ahemdabad_df["medicine"]==316]
ahemdabad_316_discount = ahemdabad_316.groupby(["discounted"]).agg({"sales":sum}).reset_index()
ahemdabad_316_discount
Out[231]:
In [232]:
ahemdabad_2103 = ahemdabad_df[ahemdabad_df["medicine"]==2103]
ahemdabad_2103_discount = ahemdabad_2103.groupby(["discounted"]).agg({"sales":sum}).reset_index()
ahemdabad_2103_discount
Out[232]:
- discount has affect on sales of medicines 316 and 2103 in ahemdabad
Lucknow Analysis¶
In [114]:
do_city_analysis(train_df_discounted_copy_city, "Lucknow")
Observations¶
- For lucknow 2017 had most sales followed by 2016 and 2018 had least sales
- For Lucknow December has the most sales followed by March, Jan and May. Aug had the least sale
- For Lucknow day 2 has most sales followed by day 3, 5, 1 and Day 31 has the least sales
Lucknow Medicine Analysis¶
In [115]:
plot_top_meds(train_df_discounted_copy_city, "Lucknow", 2015)
In [116]:
plot_top_meds(train_df_discounted_copy_city, "Lucknow", 2016)
In [117]:
plot_top_meds(train_df_discounted_copy_city, "Lucknow", 2017)
In [118]:
plot_top_meds(train_df_discounted_copy_city, "Lucknow", 2018)
In [119]:
lucknow_medicine_df = lucknow_df.groupby("medicine").agg({"sales":sum}).reset_index()
lucknow_medicine_df = lucknow_medicine_df.sort_values(by="sales", ascending = False)
lucknow_zero_sales_meds = lucknow_medicine_df[lucknow_medicine_df["sales"]==0]
lucknow_zero_sales_meds = lucknow_zero_sales_meds["medicine"]
len(lucknow_zero_sales_meds)
Out[119]:
In [120]:
lucknow_medicine_df.head()
Out[120]:
Observations¶
- For Lucknow medicine type 574 had most sales in 2015
- For Lucknow medicine type 574 had most sales in 2016
- For Lucknow medicine type 574 had most sales in 2017
- For Lucknow medicine type 2103 had most sales in 2018
- For Lucknow there are 135 medicines that had zero sales
In [233]:
lucknow_574 = lucknow_df[lucknow_df["medicine"]==574]
lucknow_574_discount = lucknow_574.groupby(["discounted"]).agg({"sales":sum}).reset_index()
lucknow_574_discount
Out[233]:
In [234]:
lucknow_2103 = lucknow_df[lucknow_df["medicine"]==2103]
lucknow_2103_discount = lucknow_2103.groupby(["discounted"]).agg({"sales":sum}).reset_index()
lucknow_2103_discount
Out[234]:
discount has affect on the sales of medicine 574 and 2103 in lucknow
Others Analysis¶
In [121]:
do_city_analysis(train_df_discounted_copy_city, "Other")
Observations¶
- for other cities 2017 has most sales followed by 2016 and 2018 has the least sales
- For Other cities March had the most sale followed by Jan , june and May. Aug had the least sales
- For Other cities day 3 has most number of sales followed by day 2, 6 and 1. Day 31 has least number of sale
For Other cities day 3 has most number of sales followed by day 2, 6 and 1. Day 31 has least number of sale
Other City Medicine Analysis¶
In [122]:
plot_top_meds(train_df_discounted_copy_city,"Other", 2015)
In [123]:
plot_top_meds(train_df_discounted_copy_city,"Other", 2016)
In [124]:
plot_top_meds(train_df_discounted_copy_city,"Other", 2017)
In [125]:
plot_top_meds(train_df_discounted_copy_city,"Other", 2018)
In [126]:
other_medicine_df = other_df.groupby("medicine").agg({"sales":sum}).reset_index()
other_medicine_df = other_medicine_df.sort_values(by="sales", ascending = False)
other_zero_sales_meds = other_medicine_df[other_medicine_df["sales"]==0]
other_zero_sales_meds = other_zero_sales_meds["medicine"]
len(other_zero_sales_meds)
Out[126]:
In [127]:
other_medicine_df.head()
Out[127]:
Observations¶
- For Other cities medicine type 574 had most sales in 2015
- For Other cities medicine type 2103 had most sales in 2016
- For Other cities medicine type 2103 had most sales in 2017
- For Other cities medicine type 2103 had most sales in 2018
- For other cities there were 186 medicines that had zero sales
Overall City Analysis¶
In [128]:
city_total_df = train_df_discounted.groupby(["city"]).agg({"sales":sum}).reset_index()
city_total_df = city_total_df.sort_values(by="sales", ascending=False)
city_total_df
Out[128]:
In [129]:
ax = sns.barplot(x="sales", y="city", data=city_total_df)
# Mumbai has highest sales
# Other cities have least sales
Observation¶
Mumbai has most sales followed by Delhi and Bengaluru
Overall Medicine Analysis¶
In [130]:
overall_medicine_df = train_df_discounted.groupby("medicine").agg({"sales":sum}).reset_index()
overall_medicine_df = overall_medicine_df.sort_values(by="sales", ascending = False)
overall_medicine_df.head()
Out[130]:
In [131]:
overall_medicine_df.tail()
Out[131]:
Observation¶
Overall medicine 1077 has higest sales followed by 2103 and 574. Also medicine 2424 has least sales
Monthly Analysis¶
As we know that overall March has the highest sales followed by January, December and August has the lowest sales. Let's see what happend in these months
In [132]:
jan_df = train_df_discounted[train_df_discounted["month"]==1]
march_df = train_df_discounted[train_df_discounted["month"]==3]
dec_df = train_df_discounted[train_df_discounted["month"]==12]
aug_df = train_df_discounted[train_df_discounted["month"]==8]
print(jan_df.shape)
print(march_df.shape)
print(dec_df.shape)
print(aug_df.shape)
In [133]:
jan_df.head()
Out[133]:
In [134]:
jan_df.tail()
Out[134]:
January Daily Analysis¶
In [135]:
jan_daily_sales_df = jan_df.groupby(["day"]).agg({"sales":sum}).reset_index()
jan_daily_sales_df = jan_daily_sales_df.sort_values(by="sales", ascending=False)
jan_daily_sales_df.head()
Out[135]:
In [136]:
jan_daily_sales_df.tail()
Out[136]:
In [137]:
ax = sns.barplot(x="day", y="sales", data=jan_daily_sales_df)
Observations¶
There are no sales on Day 1. Day 2 had highest sales followed by day 4, 3 and 5. We observe seasonality across the whole month of march
Jan Medicine Analysis¶
In [138]:
jan_medicine_df = jan_df.groupby("medicine").agg({"sales":sum}).reset_index()
jan_medicine_df = jan_medicine_df.sort_values(by="sales", ascending = False)
jan_medicine_df.head()
Out[138]:
In [139]:
jan_medicine_df.tail()
Out[139]:
In [140]:
jan_zero_sales_meds = jan_medicine_df[jan_medicine_df["sales"]==0]
jan_zero_sales_meds = jan_zero_sales_meds["medicine"]
len(jan_zero_sales_meds)
Out[140]:
Observation¶
In January medicine 1077 has most sale followed by 2103 Also there are 230 medcine that have zero sales in Jan accross, years and cities
March Daily Analysis¶
In [141]:
march_daily_sales_df = march_df.groupby(["day"]).agg({"sales":sum}).reset_index()
march_daily_sales_df = march_daily_sales_df.sort_values(by="sales", ascending=False)
march_daily_sales_df.head()
Out[141]:
In [142]:
march_daily_sales_df.tail()
Out[142]:
In [143]:
ax = sns.barplot(x="day", y="sales", data=march_daily_sales_df)
Observation¶
For March Day 1 has most sale followed by day 16, 9 , 2. Day 25 has least sale. We see a spike in sales every week
March Medicine Analysis¶
In [144]:
march_medicine_df = march_df.groupby("medicine").agg({"sales":sum}).reset_index()
march_medicine_df = march_medicine_df.sort_values(by="sales", ascending = False)
march_medicine_df.head()
Out[144]:
In [145]:
march_medicine_df.tail()
Out[145]:
In [146]:
march_zero_sales_meds = march_medicine_df[march_medicine_df["sales"]==0]
march_zero_sales_meds = march_zero_sales_meds["medicine"]
len(march_zero_sales_meds)
Out[146]:
Observation¶
In march medicine 2103 has most sale followed by 1077 Also there are 173 medcine that have zero sales in march accross, years and cities
December Daily Analysis¶
In [147]:
dec_daily_sales_df = dec_df.groupby(["day"]).agg({"sales":sum}).reset_index()
dec_daily_sales_df = dec_daily_sales_df.sort_values(by="sales", ascending=False)
dec_daily_sales_df.head()
Out[147]:
In [148]:
dec_daily_sales_df.tail()
Out[148]:
In [149]:
ax = sns.barplot(x="day", y="sales", data=dec_daily_sales_df)
Observation¶
There are no sale on Christmas day. Day 21 has most sales followed by day 23 and 22. We also see some seasonality in December
Dec Medicine Analysis¶
In [150]:
dec_medicine_df = dec_df.groupby("medicine").agg({"sales":sum}).reset_index()
dec_medicine_df = dec_medicine_df.sort_values(by="sales", ascending = False)
dec_medicine_df.head()
Out[150]:
In [151]:
dec_medicine_df.tail()
Out[151]:
In [152]:
dec_zero_sales_meds = dec_medicine_df[dec_medicine_df["sales"]==0]
dec_zero_sales_meds = dec_zero_sales_meds["medicine"]
len(dec_zero_sales_meds)
Out[152]:
Observation¶
In december medicine 412 has most sale followed by 2103 and 1077 Also there are 265 medcine that have zero sales in Dec accross, years and cities
August Daily Analysis¶
In [153]:
aug_daily_sales_df = aug_df.groupby(["day"]).agg({"sales":sum}).reset_index()
aug_daily_sales_df = aug_daily_sales_df.sort_values(by="sales", ascending=False)
aug_daily_sales_df.head()
Out[153]:
In [154]:
aug_daily_sales_df.tail()
Out[154]:
In [155]:
ax = sns.barplot(x="day", y="sales", data=aug_daily_sales_df)
Observation¶
Day 31 has most sales followed by day 2 and 30. Day 27 has least sale. We also see seasonality in august
August Medicine Analysis¶
In [156]:
aug_medicine_df = aug_df.groupby("medicine").agg({"sales":sum}).reset_index()
aug_medicine_df = aug_medicine_df.sort_values(by="sales", ascending = False)
aug_medicine_df.head()
Out[156]:
In [157]:
aug_medicine_df.tail()
Out[157]:
In [158]:
aug_zero_sales_meds = aug_medicine_df[aug_medicine_df["sales"]==0]
aug_zero_sales_meds = aug_zero_sales_meds["medicine"]
len(aug_zero_sales_meds)
Out[158]:
Observation¶
In aug medicine 1077 has most sale followed by 574 and 2103 Also there are 589 medcine that have zero sales in aug accross, years and cities
Yearly Analysis¶
In the above graphs we have seen that there is huge difference in sales in 2015 and 2016. One thing that happened in 2016 is that discounts were introduced. So, Lets deep dive
In [159]:
df_2015 = train_df_discounted[train_df_discounted["year"]==2015]
df_2016 = train_df_discounted[train_df_discounted["year"]==2016]
In [160]:
print(df_2015.shape)
print(df_2016.shape)
2015 Monthly Analysis¶
In [161]:
monthy_sales_df_2015 = df_2015.groupby(["month"]).agg({"sales":sum}).reset_index()
monthy_sales_df_2015 = monthy_sales_df_2015.sort_values(by="sales", ascending=False)
monthy_sales_df_2015
Out[161]:
In [162]:
ax = sns.barplot(x="month", y="sales", data=monthy_sales_df_2015)
Observation¶
In 2015 december has the highest sales followed by november and march and feb has the least sales. However, looking at the graph it feels that the sale never realy took off until december
2016 Monthly Analysis¶
In [163]:
monthy_sales_df_2016 = df_2016.groupby(["month"]).agg({"sales":sum}).reset_index()
monthy_sales_df_2016 = monthy_sales_df_2016.sort_values(by="sales", ascending=False)
monthy_sales_df_2016
Out[163]:
In [164]:
ax = sns.barplot(x="month", y="sales", data=monthy_sales_df_2016)
Observation¶
Both in 2015 and 2016 highest sales in December. But, the sale has almost doubled
2015 Medicine Analysis¶
In [165]:
medicine_df_2015 = df_2015.groupby("medicine").agg({"sales":sum}).reset_index()
medicine_df_2015 = medicine_df_2015.sort_values(by="sales", ascending = False)
medicine_df_2015.head()
Out[165]:
In [166]:
medicine_df_2015.tail()
Out[166]:
In [167]:
zero_sales_meds_2015 = medicine_df_2015[medicine_df_2015["sales"]==0]
zero_sales_meds_2015 = zero_sales_meds_2015["medicine"]
len(zero_sales_meds_2015)
Out[167]:
Observation¶
In 2015 medicine 1077 has most sale followed by 574 and 768 Also there are 1589 medcines that have zero sales in 2015. This implies almost half of the medicine went unsold in 2015
2016 Medicine Analysis¶
In [168]:
medicine_df_2016 = df_2016.groupby("medicine").agg({"sales":sum}).reset_index()
medicine_df_2016 = medicine_df_2016.sort_values(by="sales", ascending = False)
medicine_df_2016.head()
Out[168]:
In [169]:
medicine_df_2016.tail()
Out[169]:
In [170]:
zero_sales_meds_2016 = medicine_df_2016[medicine_df_2016["sales"]==0]
zero_sales_meds_2016 = zero_sales_meds_2016["medicine"]
len(zero_sales_meds_2016)
Out[170]:
Observation¶
In 2016 medicine 2103 has most sale followed by 1077 and 2051 Also there are 774 medcines that have zero sales in 2016. This implies number of zero sales medicines reduced by 51%
Visualizing Citywise Aggregated Sales¶
In [171]:
train_df_discounted_copy = train_df_discounted.copy()
train_df_discounted_copy.city = train_df_discounted_copy.city.astype("int")
train_df_discounted_copy.dtypes
Out[171]:
In [172]:
def convert_type_to_int(df, col_list):
for col in col_list:
df[col] = df[col].astype("int")
return df
In [173]:
train_df_discounted_copy = convert_type_to_int(df=train_df_discounted_copy, col_list=cat_columns)
In [174]:
train_df_discounted_copy.dtypes
Out[174]:
In [175]:
city = {4: 'Mumbai', 5: 'Delhi', 7: 'Bangalore', 1: 'Kolkata',
6: 'Chennai', 8: 'Hyderabad', 9: 'Ahmedabad', 3: 'Jaipur', 10: 'Lucknow', 2: 'Other'}
train_df_discounted_copy["city"]=train_df_discounted_copy["city"].map(city)
In [176]:
import plotly.express as px
In [177]:
def get_category_order_sales_for(df, col, ascending=True):
temp=df.groupby([col])["sales"].sum().sort_values().reset_index()
temp["sales"]=np.round(temp["sales"])
category_orders = list(temp[col])
return category_orders
In [178]:
get_category_order_sales_for(df=train_df_discounted_copy, col="city")
Out[178]:
In [179]:
def plot_agg_sales(df, groupby, x, color):
temp=df.groupby(groupby)["sales"].sum().reset_index()
temp["sales"]=np.round(temp["sales"])
title = f"aggregated sales for {x}"
cat_orders = get_category_order_sales_for(df, x)
fig5 = px.bar(temp, x=x, y="sales", barmode='group',
title=title, color = color,
category_orders={x:cat_orders})
fig5.show()
In [180]:
plot_agg_sales(train_df_discounted_copy, groupby = ["year", "city"], x = "city", color = "year")
In [181]:
plot_agg_sales(train_df_discounted_copy, groupby = ["month", "city"], x = "city", color = "month")
Negative Sales Analysis¶
In [182]:
train_neg_df_discounted_copy = train_neg_df_discounted.copy()
In [183]:
city_frame = {}
for city in train_neg_df_discounted_copy.city.unique():
df = train_neg_df_discounted_copy[train_neg_df_discounted_copy["city"]==city]
city_frame[city] = df
In [184]:
kolkata_neg_data = city_frame[1]
other_neg_data = city_frame[2]
jaipur_neg_data = city_frame[3]
mumbai_neg_data = city_frame[4]
delhi_neg_data = city_frame[5]
chennai_neg_data = city_frame[6]
bengaluru_neg_data = city_frame[7]
hyderabad_neg_data = city_frame[8]
ahemdabad_neg_data = city_frame[9]
lucknow_neg_data = city_frame[10]
In [185]:
city = {4: 'Mumbai', 5: 'Delhi', 7: 'Bangalore', 1: 'Kolkata',
6: 'Chennai', 8: 'Hyderabad', 9: 'Ahmedabad', 3: 'Jaipur', 10: 'Lucknow', 2: 'Other'}
train_neg_df_discounted_copy["city"]=train_neg_df_discounted_copy["city"].map(city)
train_neg_df_discounted_copy.head()
Out[185]:
In [186]:
train_neg_df_discounted_copy = change_to_category(train_neg_df_discounted_copy, cat_columns)
train_neg_df_discounted_copy.dtypes
Out[186]:
City-wise Analysis¶
Kolkata Analysis¶
In [187]:
def do_city_analysis(df, city):
city_df = df[df["city"] == city]
city_df["year"] = city_df["year"].astype("category")
city_df["month"] = city_df["month"].astype("category")
city_df["day"] = city_df["day"].astype("category")
fig_dims = (20, 20)
f, axes = plt.subplots(nrows=3, ncols=1, figsize = fig_dims)
yearly_sales_df = city_df.groupby(["year"]).agg({"sales":sum}).reset_index()
yearly_sales_df = yearly_sales_df.sort_values(by="sales", ascending=False)
sns.barplot(x="sales", y="year", data=yearly_sales_df, ax=axes[0])
monthy_sales_df = city_df.groupby(["month"]).agg({"sales":sum}).reset_index()
monthy_sales_df = monthy_sales_df.sort_values(by="sales", ascending=False)
sns.barplot(x="sales", y="month", data=monthy_sales_df, ax=axes[1])
daily_sales_df = city_df.groupby(['day']).agg({"sales":sum}).reset_index()
daily_sales_df = daily_sales_df.sort_values(by="sales", ascending = False)
sns.barplot(x="sales", y="day", data=daily_sales_df, ax=axes[2])
medicine_df = city_df.groupby(["medicine"]).agg({"sales":sum}).reset_index()
medicine_df = medicine_df.sort_values(by="sales", ascending=False)
zero_sales_meds = medicine_df[medicine_df["sales"]==0]["medicine"]
print(f"total # of medicenes with zero sale: {len(zero_sales_meds)}")
plt.show()
In [188]:
do_city_analysis(train_neg_df_discounted_copy, "Kolkata")
Observation¶
- 2015 shows least negative sales for kolkata and 2018 shows most negative sales for kolkata
- For kolkata May has the least negative sales and February has the lowest sales
- In Kolkata seventh day has the least negative sales and 2nd day has the most negative sales
Kolkata Medicine Analysis¶
In [189]:
kolkata_medicine_data = kolkata_neg_data.groupby(["medicine"]).agg({"sales":sum}).reset_index()
kolkata_medicine_data = kolkata_medicine_data.sort_values(by="sales", ascending=False)
kolkata_medicine_data.tail()
Out[189]:
Observation¶
Medicine 1515 has most number of negative sales
Jaipur Analysis¶
In [190]:
do_city_analysis(train_neg_df_discounted_copy, "Jaipur")
Observation¶
For Jaipur 2018 has the highest negative sales and 2015 has the lowest negative sales For Jaipur April has most negative sales and October has the least negative sales 3.For Jaipur Day 30 has the highest negative sales and Day 26 has the least sales
Jaipur Medicine Analysis¶
In [191]:
jaipur_medicine_data = jaipur_neg_data.groupby(["medicine"]).agg({"sales":sum}).reset_index()
jaipur_medicine_data = jaipur_medicine_data.sort_values(by="sales")
jaipur_medicine_data.head()
Out[191]:
In Jaipur medicine 2401 has the highest negative sales
Mumbai Analysis¶
In [192]:
do_city_analysis(train_neg_df_discounted_copy, "Mumbai")
For Mumbai 2017 has most negative sales and 2016 has the least sales For Mumbai December has most negative sales followed by january, June and December. August has the least negative sales For Mumbai day 15 has the highest negative sales and Day 27 has the least negative sales
Mumbai Medicine Analysis¶
In [193]:
mumbai_medicine_data = mumbai_neg_data.groupby(["medicine"]).agg({"sales":sum}).reset_index()
mumbai_medicine_data = mumbai_medicine_data.sort_values(by="sales")
mumbai_medicine_data.head()
Out[193]:
for mumbai medicine 167 has the highest negative sales
Delhi Analysis¶
In [194]:
do_city_analysis(train_neg_df_discounted_copy, "Delhi")
Delhi Medicine Analysis¶
In [195]:
delhi_medicine_data = delhi_neg_data.groupby(["medicine"]).agg({"sales":sum}).reset_index()
delhi_medicine_data = delhi_medicine_data.sort_values(by="sales")
delhi_medicine_data.head()
Out[195]:
For Delhi medicine 792 has highest negative sales
Chennai Analysis¶
In [196]:
do_city_analysis(train_neg_df_discounted_copy, "Chennai")
For Chennai 2018 has most negative sale and 2015 has least no of sales For Chennai April has the most negative sales and Aug has the least negative sales For Chennai day 24 has most negative sales and day 28 has the least Negative sales
Chennai Medicine Analysis¶
In [197]:
chennai_medicine_data = chennai_neg_data.groupby(["medicine"]).agg({"sales":sum}).reset_index()
chennai_medicine_data = chennai_medicine_data.sort_values(by="sales")
chennai_medicine_data.head()
Out[197]:
In Chennai medicine 2472 has the highest negative sale
Bengaluru Analysis¶
In [198]:
do_city_analysis(train_neg_df_discounted_copy, "Bangalore")
For Bengaluru 2017 has most negative sales and 2015 has least negative sales For Bengaluru August has most negative sales and October has least no of sales For Bengaluru Day 7 has most negative sales and day 23 has the least sales
Bengaluru Medicine Analysis¶
In [199]:
bengaluru_medicine_data = bengaluru_neg_data.groupby(["medicine"]).agg({"sales":sum}).reset_index()
bengaluru_medicine_data = bengaluru_medicine_data.sort_values(by="sales")
bengaluru_medicine_data.head()
Out[199]:
For Bengaluru medicine 1098 has the highest negative sales
Hyderabad Analysis¶
In [200]:
do_city_analysis(train_neg_df_discounted_copy, "Hyderabad")
For Hyderabad 2017 show most negative sales and 2015 has least sales For Hyderabad May has most negative sales July has least negative sales For Hyderabad Day 30 has the most negative sale and day 17, 23, 9 has the least sales
Hyderabad Medicine Analysis¶
In [201]:
hyderabad_medicine_data = hyderabad_neg_data.groupby(["medicine"]).agg({"sales":sum}).reset_index()
hyderabad_medicine_data = hyderabad_medicine_data.sort_values(by="sales")
hyderabad_medicine_data.head()
Out[201]:
For Hyderabad medicine 846 has the highest negative sales
Ahmedabad Analysis¶
In [202]:
do_city_analysis(train_neg_df_discounted_copy, "Ahmedabad")
- for Ahemdabad 2017 had most negative sales and 2016 had least negative sales
- For Ahemdabad July has most negative sales followed by March, June and Jan. sept has the least negative sales
- For Ahemdabad day 20 had most negative sales and day 20 has least sales
Ahmedabad Medicine Analysis¶
In [203]:
ahemdabad_medicine_data = ahemdabad_neg_data.groupby(["medicine"]).agg({"sales":sum}).reset_index()
ahemdabad_medicine_data = ahemdabad_medicine_data.sort_values(by="sales")
ahemdabad_medicine_data.head()
Out[203]:
In ahemdabad medicine 1144 has highest negative
Lucknow Analysis¶
In [204]:
do_city_analysis(train_neg_df_discounted_copy, "Lucknow")
In Lucknow medicine 3194 has the highest negative sales
Others Analysis¶
In [205]:
do_city_analysis(train_neg_df_discounted_copy, "Other")
For other cities 2018 has most negative sales and 2015 has the least sales For Other cities April had the most negative. Aug had the least negative sales For Other cities day 23 has most number of negative sales Day 3 has least negative sale
Other City Medicine Analysis¶
In [206]:
other_medicine_data = other_neg_data.groupby("medicine").agg({"sales":sum}).reset_index()
other_medicine_data = other_medicine_data.sort_values(by="sales")
other_medicine_data.head()
Out[206]:
For other cities medicine 1762 has highest negative sales
Solution Design Overview¶
In [ ]:
In [ ]:
In [ ]: